A complete, autonomous and parallel photonic neural network in a semiconductor laser
Résumé
We
experimentally demonstrate a fully parallel photonic neural network using
spatially distributed modes of a large-area vertical cavity surface-emitting
laser (LA-VCSEL). All components of the ANN are fully realized in parallel
hardware: 100+ nodes are implemented in a LA-VCSEL, input and output weights
are realized via the complex transmission matrix of a multimode fibre and a
digital micro-mirror array, respectively. We train the readout weights to
perform header recognition (up to 6 bit), XOR classification and digital to
analog conversion, and obtain low error rates for all tasks. Our system uses
readily available components, is scalable to much larger sizes and to
bandwidths in excess of 20 GHz. We also explore more efficient training
strategies and different optical weight initializations to increase
performance.